Milkisa Yebasse

About Me

I am an applied AI researcher with a diverse academic background, having completed my bachelor's degree in Ethiopia, my master's degree in South Korea, and currently pursuing a PhD in Italy. This journey has provided me with a broad, international perspective. My research focuses on applying AI across various domains, including radar image analysis, medical image analysis, and crop disease detections. I am passionate about using AI to extract meaningful insights from complex datasets and explore innovative solutions in these fields. More broadly, I am dedicated to applying AI to solve real-world problems, particularly in addressing climate change and planetary exploration, while promoting AI ethics and ensuring that AI is accessible to all.

Name: Milkisa Tesfaye
Nationality: Ethiopia
Phone: +393509689235
City: Trento, Italy

Age: 27
Degree: Computer Engienering
Email: milkisatesfaye@gmail.com
Job: Ph.D Student

Certificate Coursework's in machine learning and computer vision.

Years of Teaching Experience My experiemce working as an assistant lecturer ata Hawassa University and Madda Walabu University has been rewarding in terms of my teaching and leadership experience

Years of Research experienceAs a research assistant during my master’s and PhD program, I collaborated with lab members to accomplish research goals. I also carried out independent research and produced a paper.

Awards and publication I have recieved awards such as Global Korea Scholarship, Tensorflw Developer,award for university promotional video and i published a paper on coffee disease visualizatoin and classification.

Python 100%

Java 90%

C++75%

My Resume

Learning is a Lifelong Process

Education

Ph.D in information engineering and computer science

2022 - 2025

University of Trento, Italy

Supervisor: Prof. Lorenzo Bruzzone
Research Focus: Application of AI for planetary radars.

Master of Engineering in Computer Engineering

2020 - 2022

Kumoh National Institute of Technology, South Korea

Supervisor: Prof. Jaepil Jo
GPA: 4.44/4.5
Research Focus: Medical image analysis using AI.
I studied courses such as special topics in AI, Big data, multimodal interaction, probability and Random Process.

Bachelor of Science in computer Engineering

2012 - 2017

Hawassa University, Ethiopia

GPA: 3.61/4.0

Certification

Certified in Deep learning Specialization, DeepLearning.AI on Coursera.

Completed courses: Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization, Neural Networks and Deep Learning, Sequence Models, Structuring Machine Learning Projects, Convolutional neural network.

DeepLearning.AI Tensorflow Developer Professional Certificate.

Handle real-world image data and explore strategies to prevent overfitting, Build natural language processing systems using tensorflow and apply RNN,GRUs, and LSTMs as you train them using text repositories.

Certified in Advanced Computer Vision from DeepLearning.AI on Coursera.

Practiced complex object detection, image segmentation, and visual interpretation of convolutions.

Deep Neural Networks with PyTorch.

This course cover different models starting off with fundamentals such as Linear Regression, and logistic/softmax regression. Followed by Feedforward deep neural networks, the role of different activation functions, normalization and dropout layers. Then Convolutional Neural Networks and Transfer learning is also covered using PyTorch.

Research and Teaching Experience

Doctorial Research Assistant

10-2022 - Present

Remote sensing labratory, Trento, Italy

Performing deep learning-based automatic analysis of radar sounder data collected from past missions, including MCoRDS and planetary datasets acquired by SHARAD from the polar regions of Earth and Mars, respectively..
Developing automatic analysis techniques with limited labels, such as few-shot learning and semi-supervised learning methods, to efficiently manage the data volume of future surveys and effectively extract valuable information.
Developing a GUI application for the automatic labeling of radar sounder data.

Masters Research Assistant

09/2020 - 09/2022

computer vision and pattern recognition labratory,Kumoh National Institute of technology, South Korea

Participated on paper reviws and project proposal writing.
Conducted experimentation on coffee disease classification experiments and published a paper.

Teaching Assistant

12/2018 - 08/2019

Hawassa University,Ethiopia

Participated in weekly faculty meetings and contributed to curriculum review.
Mentored undergraduate students for their theses.
Taught courses such as computational and numerical methods and C++.

Teaching Assistant

08/2017 - 08/2019

Madda Walabu University,Ethiopia

Wrote course materials such as course outline, homework assignments and handouts.
Mentored undergraduate students for their theses.
Taught courses such as Programming Language and C++.

Honors and Award

Global Korea Scholarship.

National Institute for International Education under the Ministry of Education awarded me as a scholar of 2019 Global Korea Scholarship Program for Graduate Degrees. NIIED sponsored my studies in Korea throughout my master’s degree. The scholarship benefits include a round-trip air fare, tuition, monthly allowance, and insurance fees.

Tensorflow Developer Certificate, Tensorflow.

Passed the tensorflow developer certificate exam covering foundational, practical machine learning skills through the building and tracking of models using tensorflow.

Award for University Promotional Video.

I received an award from ministry of education of The republic of korea in recognition to the university promotion video i made.

Graduated with Distinction.

I graduated from Hawassa University on July 29, 2017, with a BSc Degree in Computer Engineering with distinction.

Journal Publication

Semi-Supervised Semantic Segmentation of Radar Sounder Data with Sparse Annotations. (In preparation)

In this paper,we propose a semi-supervised learning method that leverages sparse annotations, including points, horizontal lines, diagonal lines, and polygons, alongside pseudo-labels generated from unlabeled data.

A Spatially Aware Few-shot Approach to Classification of Radar Sounder Data.

In this paper,we propose a novel few-shot pixel-based classification framework for radar sounder data. This framework aims to learn underlying patterns using only a few labeled support samples and adapt quickly to radargrams from different campaigns with minimal labeled information without requiring retraining.

Malaria Disease Cell Classification With Highlighting Small Infected Regions.

In this paper,we propose a simple neural network training strategy for highlighting the infected pixel regions that are mainly responsible for malaria cell classification.

Coffee Disease Visualizatoin and classification.

In this paper, We present a guided approach that achieved 98% accuracy in coffee disease classification. Further, in this study, we provide visualization of coffee disease, which exclusively highlights the region responsible for classification.

Conference Publication

Scribble-Driven Semi-Supervised Semantic Segmentation of Radar Sounder Data. (Under review)

Venue:International Geoscience and Remote Sensing Symposium (IGARSS).(Oral).

This paper introduced a semi-supervised method for radar sounder data semantic segmentation based on scribble annotations. Our approach optimizes scribble-guided KNN for label propagation and utilizes a multi-task efficient u2net architecture to effectively address the challenges posed by inter-class similarities and ambiguous boundaries.